Functional dissection of the Drosophila melanogaster Fibroblast Growth Factor signalling pathway in branching morphogenesis
of the developing tracheal system
Inauguraldissertation
zur
Erlangung der Würde eines Doktors vorgelegt der
Philosophisch-Naturwissenschaftlichen Fakultät der Universität Basel
von
Caroline Dossenbach
von Baar/ ZG
Ausgeführt unter der Leitung von Prof. Markus Affolter
Abteilung Zellbiologie Biozentrum der Universität Basel
Klingelbergstrasse 50-70 CH-4056 Basel
2004
auf Antrag von:
Prof. Dr. Markus Affolter Prof. Dr. Walter J. Gehring Prof. Dr. Silvia Arber (Dissertationsleiter) (Korreferent) (Vorsitzende)
Basel, den 04. 05. 2004
Prof. Dr. Marcel Tanner (Dekan)
Index
INDEX ... 3
ACKNOWLEDGEMENTS... 6
SUMMARY... 8
INTRODUCTION FGF SIGNALLING ... 11
I. CHARACTERISATION OF THE FGFRS AND DOWNSTREAM SIGNALLING CASCADES... 13
1. The extracellular domain of FGFRs ... 14
2. The intracellular domain of FGFRs ... 15
3. Different cellular responses elicited by FGFR1... 17
3.1The MAPK cascade... 17
3.2The adapter FGFR substrate 2 (FRS2) ... 17
3.3The mitogenic response ... 18
3.4The cell survival response... 19
3.5The migratory response ... 20
4. Negative regulation of FGF signaling... 22
4.1The ubiquitin ligase Cbl ... 22
4.2MAP kinase-mediated negative feedback mechanism ... 23
4.3PLCγ as negative regulator of FGFR... 23
4.4Sprouty (Spry) ... 23
4.5Sef (similar expression of fgf genes) ... 25
II. CHEMOTACTIC CELL MOVEMENT... 26
1. Molecular requirements for actin-based lamellipodia and filopodia formation ... 29
2. Role of the small Rho GTPases in cytoskeletal reorganization ... 31
III. DEVELOPMENTAL ROLE OF FGF SIGNALLING IN VERTEBRATES... 34
1. Tumorogenesis... 34
2. FGF-mediated movement of mesoderm cells during gastrulation ... 34
3. Angiogenesis... 36
4. Development of the mammalian lung... 38
4.1The morphology of the lung... 38
4.2Cell fate determination ... 39
4.3The branching process... 39
IV. DROSOPHILA MELANOGASTER AS MODEL ORGANISM TO STUDY FGF-MEDIATED CHEMOTAXIS AND CELL MOTILITY42 1. FGF signalling pathway in Drosophila... 42
1.1Sugarless (Sgl) and Sulfatless (Sfl)... 42
1.2Breathless (Btl) and Heartless (Htl) ... 43
1.3Downstream-of-FGFR (Dof) ... 44
1.4Corkscrew (Csw) ... 46
2. Developmental role of FGF signalling in Drosophila... 48
2.1Htl and its role during gastrulation ... 48
2.3The tracheal system of Drosophila melanogaster... 51
a) Tracheal cell fate determination... 52
b) Invagination and primary branch outgrowth... 53
c) FGF signalling acts as guidance cue and induces directed cell migration ... 54
d) Secondary and terminal branch formation ... 55
2.4FGF as chemoattractant of the air sacs ... 56
AIM OF THE THESIS... 58
EXPERIMENTS: ... 60
FUNCTIONAL DISSECTION OF THE FGF SIGNALLING PATHWAY... 60
I. THE GAL4-UAS SYSTEM FOR DIRECTED GENE MISEXPRESSION... 60
II. LOCALIZATION OF BREATHLESS IN THE OUTGROWING TRACHEAL BRANCHES... 62
III. PUBLICATION:SPECIFICITY OF FGF SIGNALLING IN CELL MIGRATION IN DROSOPHILA... 64
IV. PUBLICATION:DOWNSTREAM-OF-FGFR(DOF) IS A FGF-SPECIFIC SCAFFOLDING PROTEIN AND RECRUITS CORKSCREW UPON RECEPTOR ACTIVATION... 65
V. FUNCTIONAL DELETION ANALYSIS OF BREATHLESS (BTL) IN VIVO AND IN VITRO... 66
1. Constructs ... 66
2. Rescue of tracheal cell migration in btl mutant embryos by the truncated Btl receptors ... 70
3. S2 assay: Molecular analysis of the Btl-Dof interaction ... 72
VI. THE ROLE OF DRAS,DRAF AND ROLLED IN THE MIGRATORY RESPONSE... 78
1. Ras effector mutants ... 78
2. Local activation of Raf ... 82
3. MAPK activity in tracheal cells ... 84
VII.THE RHO FAMILY OF SMALL GTPASES... 89
1. Cdc42 heteroallelic mutant embryos... 93
2. Rac germline clones... 94
VIII. COMPARISON TO VERTEBRATE FGF SIGNALING... 97
IX. BTL MUTANT MOSAIC CLONES IN THE TRACHEAL SYSTEM... 103
DISCUSSION ... 107
I. TRACHEAL AND MESODERMAL CELLS RESPOND TO FUNCTIONALLY DISTINCT RTKS WITH DIRECTED MIGRATION107 II. DOF ACTS SPECIFICALLY IN THE FGF SIGNALLING PATHWAY... 108
1. Dof provides a functional homologue of FRS2 in Drosophila ... 109
2. Dof probably interacts constitutively with the kinase domain of Btl... 111
III. CSW PROVIDES THE ESSENTIAL LINK TO THE CYTOSKELETON... 112
V. MITOTIC CLONES TO IDENTIFY NEW TARGETS INVOLVED IN CELL MIGRATION... 117
1. Single mutant tracheal cells ... 117
2. Larval air sac development is promising to identify regulators of cell migration ... 118
3. Single cell rescue experiment using the flipout system ... 120
VI. CONCLUDING REMARKS... 121
MATERIALS AND METHODS ... 123
I. DROSOPHILA STRAINS AND GENETICS... 123
II. RESCUE ASSAYS... 124
1. Schematic representation of the crosses ... 125
III) 126 III. GENETIC MOSAIC ANALYSIS USING THE FLP/FRT SYSTEM... 127
IV. RAC GERMLINE CLONES... 128
V. IMMUNOHISTOCHEMISTRY... 129
1. Embryo collection and fixation... 129
2. Protocol of fluorescent antibody stainings... 129
3. Antibodies... 130
VI. MOLECULAR BIOLOGY... 131
1. Cloning ... 131
1.1Btl derivates to generate transgenic flies... 131
1.2Btl derivates to transfect S2 cells ... 133
2. The polymerase chain reaction (PCR) ... 135
3. Restriction digests ... 136
4. Ligation and bacterial transformation ... 136
VII.S2 CELL CULTURE ASSAY... 136
1. Cell transfection... 136
2. Cell lysates ... 137
3. Immunoprecipitation ... 137
4. Western blots... 137
REFERENCES... 138
APPENDIX ... 147
I. PUBLICATIONS... 148
II. PARTICIPATION AT FOLLOWING CONGRESSES... 149
III. TUTORIAL RESPONSIBILITIES... 149
IV. LECTURES... 150
V. EID... 151
Acknowledgements
First, I would like to thank Prof. Markus Affolter for giving me the opportunity to do the thesis in his lab. I thank him for his precious support, his continuous encouragement and guidance through my projects. His enthusiasm about Developmental Biology and Science in general was very stimulating and motivating. Thank you very much for everything!
I would like to thank Prof. Walter J. Gehring and Prof. Silvia Arber for being referees in my thesis committee.
Additionally, I thank Prof. Gehring for his support. Furthermore, he broaded my horizon by providing me the opportunity to join the great marine biology course he organized in Banyuls. This course introduced me into the fascinating world of the Ocean and Developmental Biology.
Prof. Anna Seelig and Prof. Christian Rehmann-Sutter also enriched my horizon by reflecting the world of Science from the Ethical and Philosophical point of views. Especially, I would like to Prof. Seelig for her support and ideas concerning my future work.
Special thanks go to all the members of the lab: for their help, advices, discussions (scientific and non-scientific ones) and for providing a such a nice atmosphere in the lab.
Ute Nussbaumer taught me a lot of techniques and was always very helpful. I enjoyed spending time with her, and I am thankful for her friendship. Thomas Marty introduced me enthusiastically into the world Drosophila and taught me how to do Genetics. Valérie Petit helped me with a lot of technical but also scientific advices concerning the projects we shared. I enjoyed the collaboration and discussions with Marc Neumann and his fruitful advices. Especially, I would like to thank him for taking time to read and comment my whole thesis manuscript. I would like to thank Anja Jazwinska for teaching me Genetics and how to generate germline clones. And for the great time, I spent with Anja in Chicago during the “Annual Drosophila Research Conference”, but also as tourists visiting this spectacular town.
I would like to thank the other lab members for all their help and advices: Jorgos Pyrowolakis, Britta Hartmann, Clemens Cabernard, Samir Merabet, Alain Jung, Carlos Ribeiro and Andreas Ebner. With Andy I spent a good time in the corner of the lab, and we shared a lot of funny moments. I also enjoyed the discussions with the diploma students Christina Nef, Sven Falk and Andreas Stadler.
I also thank all members of the second floor of the Biozentrum for providing a nice atmosphere. Especially Lydia Michaut and Véronique Charpignon for sharing my enthusiasm for dancing. I would like to thank Tomoko Nagao for giving me the opportunity to learn more about the interesting culture of Japan after the “International Congress of Developmental Biology” in Kyoto.
I thank Markus Dürrenberger and Carlos Ribeiro for teaching me using the Confocal microscopy. I thank Liliane Devaja and Greta Backhaus for their secretarial assistance. I thank the people from BioPhit, who helped immediately with Computer problems. Karin Mauro, Bernadette Bruno and Gina Evora supported our scientific work by supplying fly food, buffers and by cleaning all the lab goods.
I would like to thank the fly community to share research material. Many thanks also to Kathy Mathews, the Bloomington Drosophila Stock Centre at the Indiana University and the Fly Base team.
Many thanks go to my friend Elisa Lepori for comments on my thesis manuscript and for all the precious moments outside the lab.
Also many thanks to Rebecca Peter and to the URZ for lending me good laptops on which this thesis was written.
Finally, I would like to thank my family for loving and supporting me through my whole life. Words fail me to describe the gratitude and love I feel for Fabian Möller.
Summary
Fibroblast Growth Factor (FGF) signalling is involved in numerous developmental processes ranging from cell determination to mitogenesis, and cell survival to cell migration. Interestingly, the same signalling pathway is used reiteratively throughout development and the question regarding the intracellular specificity is raised.
Little is known about the intracellular signalling events of the FGF signalling pathway leading to specific cellular responses. Since the FGF signal is essential throughout embryonic and adult development and plays a role in many pathogenical processes, it is important to identify the factors, which determine the differential responses.
We were interested to investigate the specificity of FGF signalling in a developmental context in which the signal induces directed cell migration, a cellular phenomenon that relies on changes of the cytoskeletal architecture.
During gastrulation in early embryonic development, but also during the formation of organs in mammals and in Drosophila, FGFs have been shown to act as chemoattractants and guide cells toward their targets. In these contexts, FGF signalling has been shown to induce filopodia, which are long cellular extensions containing parallel actin bundles.
Using Drosophila tracheal and mesodermal cell migration as model systems, we found that the intracellular domain of the two Drosophila FGF receptors (FGFRs) Breathless (Btl) and Heartless (Htl) can be replaced by the equivalent domains of Torso and EGFR, and yet these hybrid receptors will rescue cell migration in btl or htl mutant embryos, respectively. These chimeric receptors rescued cell migration even in the absence of Downstream-of-FGFR (Dof), a scaffolding protein that has been shown to be essential for FGF signalling in Drosophila. Thus, Dof acts specifically in the FGF signalling pathway. The functional characterization of Dof has demonstrated that Dof is indeed a FGFR specific phosphotarget and forms a complex with both FGFRs, but it is not a substrate of Torso.
We performed a functional deletion analysis of the Btl receptor to define the interaction domains of Dof and other putative adapter proteins essential in the process of cell migration. Deletion of all putative interaction domains outside of the kinase domain did not affect the rescue capacity of the truncated Btl receptors in vivo suggesting that the kinase domain is sufficient for transmitting the signal. In line with this interpretation, results from S2 cell culture experiments revealed that Dof interacts with the kinase domain, and it does so independently of the activation state of the receptor. Surprisingly, in S2 cells, Btl receptors lacking the C-terminus did not auto-phosphorylate, as consequence we could not observe phosphorylation of Dof. We assume, that the short C- terminus is required for conformational changes of the kinase activation loop upon dimerization of the receptors to enable trans-phosphorylation.
Dof belongs to a distinct family of adapter proteins than its functional homologue, the vertebrate FGFR adapter protein FRS2 that has been shown to constitutively interact with the juxtamembrane domain of the FGFRs. We could show that the human FGFR2, when expressed in the tracheal system, is only able to rescue cell migration defects effectively in the presence of Dof. These results suggest that Dof is able to interact with human FGFRs. At present, there is no evidence for a FRS2 homologue in Drosophila that might act as substitute for Dof.
Upon receptor activation, Dof recruits the phosphatase Corkscrew (Csw), the Drosophila Shp2 homologue. Csw recruitment represents an essential step in FGF induced cell migration and transcriptional activation via the Ras/MAPK cascade. However, our results indicate that the activation of Ras is not sufficient to activate the migration machinery in tracheal and mesodermal cells. Ectopic activation of the Ras/MAPK cascade partially rescued tracheal cell migration in btl or dof mutant embryos. But high levels of sustained activation of Ras or the MAPK in wild-type tracheal cells did not disturb the migratory behaviour of the cells in contrast to ectopic activation of Branchless (Bnl), the Drosophila FGF homologue, which completely impaired primary branch outgrowth. In a wild-type tracheal system, MAPK activity is restricted to the tracheal tip cells. Single cell rescue experiments indicate that Bnl induces the migratory response
exclusively in the tip cells of the outgrowing tracheal branches; the stalk cells are pulled forward by cell-cell adhesion contacts.
The small GTPases of the Rho family have been shown to regulate cytoskeletal rearrangements. For tracheal development, Dcdc42 could function in collaboration with Drac in the regulation of actin dynamics according to our experiments. Additional proteins linking either Dof or Csw to the small GTPases have to be identified.
Introduction FGF signalling
The Fibroblast Growth Factors (FGFs) constitute a large family of proteins that act in a wide range of developmental and pathological processes. The biological functions of FGFs include the regulation of cell proliferation, differentiation, survival, motility and tissue patterning.
In early and late embryogenesis, FGF signalling is essential for mesoderm induction, midbrain development and patterning of the neural plate, proper development and branching of the lung, the development of the skin and the growth and patterning of the limbs and skeletogenesis.
In the adult, FGFs are involved in tissue repair, wound healing and neurite outgrowth, as well as in the maintenance of the vasculature system and the migration in angiogenesis.
By their spatially and temporally restricted availability, FGFs act as ligands at the outside of the cells ensuring correct activation of the FGF receptors (FGFRs) that belong to the receptor tyrosine kinase (RTK) family. However, FGF signalling is reiteratively used and intracellular signalling components are highly conserved within the RTK signalling. The question arises, how intracellular specificity is achieved resulting in all the distinct cellular responses.
Why is the same signalling pathway interpreted differently in different tissues?
One model proposes intrinsic differences in the intracellular signalling pathway.
A second model suggests that specificity arises from differences in the magnitude or duration of MAPK activation, one of the conserved components of RTK signalling (Halfar et al., 2001; Marshall, 1995). And finally, a third model postulates that RTKs generally act via the same signalling cassette, producing a generic signal, but cells interpret these signals according to their distinct developmental histories (Simon, 2000).
This introduction focuses on one specific cellular response to FGF signalling, namely the induction of cell motility during distinct processes of embryonic development.
Diverse migratory processes of vertebrate embryonic development are compared to the fruit fly Drosophila melanogaster, a model organism to study the basic strategies underlying development. Most developmental processes have been conserved among multicellular organisms.
Furthermore, the introduction gives an overview about the molecular players and regulators in the FGF signalling pathway acting upstream and downstream of the FGFR in Vertebrates and in Drosophila.
I. Characterisation of the FGFRs and downstream signalling cascades
In human and mice, 23 FGFs exist which are ca. 30-70% identical in their primary amino acid sequences and are proteolytically processed. So far 4 FGFs have been identified in zebrafish, 7 in Xenopus and 7 in chicken. In invertebrates only one FGF homologue Branchless (Bnl) has been identified in Drosophila and two in Caenorhabditis elegans (Egl17 and Let756) (Ornitz and Itoh, 2001).
Most FGFs have amino-terminal signal peptides and are secreted from the cells. Some FGFs are not secreted, but however, are found on the cell surface within the extracellular matrix (FGF1 or acidic FGF, FGF2 or basic FGF). Some remain intracellular because they lack the classical signal sequence that allows efficient export from the cell. FGF1 stimulates in vitro endothelial cell migration and proliferation. FGF2 exists as a cytoplasmic 18-kDa isoform and four high molecular weight (HMW) isoforms that are nuclear based. A new identified protein called translokin, which is associated with the microtubular network, specifically binds FGF2. Mutation of the nuclear targeting sequence of translokin or RNAi interference abrogates the intracellular translocation of FGF2 (Auguste et al., 2003). HMW isoforms but not the 18-kDa isoform of FGF2 have a N-terminal sequence responsible for the nuclear targeting/retention signal.
Dominant negative strategies in cultured cells demonstrated that HMW FGF stimulate DNA synthesis independent of cell surface receptors to induce the mitogenic response, but the exact role of intracellular FGF still has to be elucidated (Auguste et al., 2003; Ornitz and Itoh, 2001; Powers et al., 2000).
In vertebrates, 4 structurally related FGFR exist and alternative splicing results in even more receptor variations.
The expression patterns of the FGFR are distinct but overlapping during embryonic development. In the mouse embryo FGFR1 is expressed in the mesenchyme and predominantly in the brain, FGFR2 in several epithelial
tissues, FGFR3 predominantly in the brain, spinal chord and cartilage rudiments of the developing bone and FGFR4 in tissues of endodermal and mesodermal origin (Klint and Claesson-Welsh, 1999).
1. The extracellular domain of FGFRs
All receptors consist of an extracellular domain with 3 Immunoglobulin-like domains (Ig-like domains), a stretch of acidic amino acids between Ig-like domain 1 and 2 that is unique for FGFRs and a heparin-binding domain.
FGFR1-3 have different splice variants, these receptors either have two or three Ig-like domains. Two Ig domain variants may or may not have the acidic box and the receptors may even lack the transmembrane and intracellular parts. A splice variant of FGFR3 for example, in which the transmembrane domain is deleted, was reported to be localised into the nucleus (Klingelberg Diss). The third Ig-like domain can be spliced into IIIb or IIIc isoforms that have different ligand binding specificity. Ligand binding and specificity resides in IgII and IgIII and the linker that connects them (Plotnikov et al., 2000).
Heparan sulfate proteoglycans (HSPGs) such as syndecan and glypican are a class of molecules that act as coreceptors of FGFRs and stabilise ligand- receptor complex formation. Proteoglycans are abundant components of the extracellular matrix and are composed of core proteins with covalently attached modified glycosaminoglycan (GAG) polysaccharide polymer side chains. All heparan GAG chains undergo some N-deacetylation/N-sulfation and O-sulfation and they are referred as heparan sulfate (HS). Sulfation is responsible for the majority of the structural diversity of HS chains and different FGFs have different affinities for unique HS sequences. This suggests that tissue specific pattern of O-sulfation and the local concentration of HS can regulate the activity and specificity of FGFs (Nybakken and Perrimon, 2002; Ornitz, 2000).
Crystallography experiments have shown that a minimal complex of FGF2/FGFR1 in the absence of heparin/HS is formed (Plotnikov et al., 2000;
Plotnikov et al., 1999). This minimal complex allows transient receptor dimerization and may signal at high ligand concentrations. Heparin/HS and a second FGF molecule are required to stabilise the active complex to gain a
sustained specific intracellular response. One model suggests that heparin interacts with both FGF and FGFR and promotes the formation of a stable 1:1:1 FGF:FGFR:heparin ternary complex. A second ternary complex is then recruited to the first complex via direct FGFR:FGFR contacts, secondary interactions between FGF in one ternary complex and FGFR in the other complex, and indirect heparin-mediated FGFR-FGFR contacts. There are no direct FGF-FGF interactions. One role of the acidic box laying between IgI and IgII could be that it competes for the binding of heparin to the basic heparin binding domain of FGFR thereby providing an autoinhibitory effect (Plotnikov et al., 1999).
In lazy mesoderm (lzme) mutant mice, the UDP-glucose dehydrogenase, an enzyme required for the synthesis of the glycosaminoglycan side chains of the proteoglycans is not produced. In this mutant, GAG production is blocked and FGF signalling is disrupted downstream of FGF ligand expression since expression of FGF signalling target genes is reduced as for example mspry2 comparable to FGF8-/- embryos (Garcia-Garcia and Anderson, 2003).
2. The intracellular domain of FGFRs
FGFRs have a transmembrane domain followed by the intracellular part that consists of a long juxtamembrane domain, a split tyrosine kinase domain with an N-terminal lobe and a C-terminal lobe and a short C-Terminus.
Binding of two FGFs lead to receptor dimerization, conformational changes and autophosphorylation of the receptors on 7 conserved tyrosines.
The kinase domain contains two functionally important binding sites, the ATP binding site and the substrate binding site. In the unphosphorylated FGFR1 the ATP binding site is accessible while the substrate binding site is blocked in the so-called activation loop. When another FGFR molecule comes close, the activation loop changes its conformation, the two conserved tyrosine residues in the activation loop (Y653 and Y654) get trans-phosphorylated thereby upregulating the kinase activity and the phosphorylation of the other 5 phosphotyrosine residues (Klingenberg Dissertation). Two tyrosine residues Tyr
677 and 701 in the C-terminal lobe are not sites of autophosphorylation but they play a role in the stabilisation of the activation loop (Foehr et al., 2001).
The remaining 3 phosphotyrosines are potential binding sites for Src homology 2 (SH2) or phosphotyrosine binding domain (PTB) containing adapter proteins.
SH2 domains are 100 aa residue long conserved motifs that are found in different enzymes or in proteins that lack enzymatic activities denoted as adapters. SH2 containing proteins are affected by binding to the receptor, either by tyrosine phosphorylation, and conformational changes and by increased availability of their substrates giving rise to a cellular response. The same is true for PTB domain proteins that contain the 100-150aa conserved domain.
In the juxtamembrane domain of FGFR1 and FGFR2, there is one phosphorylated tyrosine motif (Y463) that serves as binding site of the adapter Crk (FGFR3 and FGFR4 lack the tyrosine in the juxtamembrane domain). Crk binds the guanine nucleotide exchange factor C3G that has been shown to activate the monomeric GTPase Rap1 to mediate sustained activation of the Mitogen-activated protein kinase (MAPK) ERK1 and 2.
Two phosphorylated tyrosines are in the kinase insert domain of FGFR1 and 2 (Y583, Y585), one in FGFR3 and none in FGFR4. The kinase insert tyrosine residues appear to be dispensable for FGFR1. Finally, the C-terminus (Y766) serves as binding site for PLCγ and the adapter Shb (Klint and Claesson-Welsh, 1999). Binding of PLCγ to the activated FGFR1 leads to its translocation to the plasmamembrane and subsequent phosphorylation upon FGFR1. At the plasmamembrane, PLCγ catalyses the hydrolysis of phosphoinositol lipids to inositol phosphates and diacylglycerol, which in turn stimulate the release of intracellular Ca2+ and activation of the protein kinase C (PKC).
Experiments have shown that PLCγ-activation is neither required for FGFR1 mediated mitogenesis, nor for chemotaxis nor for neurite outgrowth (Klint and Claesson-Welsh, 1999).
Shb is an ubiquitously expressed adapter protein containing SH2 and PTB domains. Interestingly, Shb binds to the same tyrosine motif at the C-terminus as PLCγ and according to experiments they don’t compete for each other (Cross et al., 2002).
3. Different cellular responses elicited by FGFR1
The FGFR1 is the best-characterised FGFR and intensive studies about the interaction with downstream signalling components have been done.
3.1 The MAPK cascade
In contrast to other RTKs, all FGFRs lack a Grb2 binding site. The adapter Grb2 is constitutively bound to the guanine nucleotide exchange factor (GEF) Son of Sevenless (SOS). SOS catalyses the exchange of GDP for GTP on Ras, thereby activating Ras which is tethered to the membrane by farnesylation.
GTP-bound Ras binds the N-terminal domain of the serine/threonine kinase Raf targeting Raf to the membrane which is sufficient for its activation. Raf then phosphorylates and activates the dual specificity kinase MEK that finally activates and phosphorylates the MAPK, ERK1 and ERK2. Oligomers of the activated and phosphorylated MAPK ERK1 and 2 enter the nucleus and phosphorylate transcription factors thereby regulating the expression of certain genes for a given cellular response.
3.2 The adapter FGFR substrate 2 (FRS2)
In vertebrate FGF signalling, Grb2 is recruited to the FGFR by FRS2 (FGF receptor substrate 2, also called suc1-associated neurotrophic factor target, SNT-1) (Kouhara et al., 1997). FRS2 is a membrane associated (myristoylated) multiadapter protein that binds constitutively to a stretch of the juxtamembrane domain of the FGFR1 (407-433aa) with its N-terminal PTB domain in a phosphotyrosine independent interaction. PTB domains usually interact with the canonical NPXY motif or asparagine residues, but the juxtamembrane domain lacks this motif or asparagine residues. That is unique and it was suggested that FRS2 transiently contacts the FGFR prior to receptor activation (Ong et al., 2000a; Xu et al., 1998).
Nuclear magnetic resonance (NMR) studies have shown that the PTB domain of FRS2 possesses unique features. The most important difference is that the PTB domain contains an additional β-strand (β8) that is essential for the
interaction with the conserved domain of the juxtamembrane domain of FGFR1 (Dhalluin et al., 2000). At the C-terminus, FRS2 contains multiple tyrosine phosphorylation domains, among those 4 Grb2 binding sites and two binding sites for the protein tyrosine phosphatase Shp2 that carries an additional Grb2 binding site. Thus, Grb2/SOS complexes are recruited either directly or indirectly via Shp2 upon tyrosine phosphorylation of FRS2 in response to FGF stimulation. The C-terminus of FRS2 is required for the recruitment of SOS either to contact SOS directly or through intermediate proteins (Hadari et al., 2001; Xu and Goldfarb, 2001; Xu et al., 1998).
Two isoforms of FRS2 exist, they are highly homologous, but they have a different expression pattern. FRS2α is expressed ubiquitously and can be detected at every developmental stage of the mouse, whereas the expression of FRS2β begins at day 9 and is primarily confined to tissues of neuronal origin.
In contrast to FRS2α, FRS2β does not bind constitutively to the juxtamembrane domain of FGFRs but only to activated TrkA receptors to a canonical tyrosine phosphorylated NPXY motif (Lax et al., 2002).
3.3 The mitogenic response
FGFs are potent inducers of DNA synthesis in multiple cell types. For the mitogenic response, complex formation of the two adapters FRS2 at the juxtamembrane domain, Shb at the C-terminus and the phosphatase Shp2 is required for maximal and sustained activation of the MAPK. Shb probably directly regulates phosphorylation and association of Shp2 with FRS2. The mechanism whereby Shp2 regulates FRS2 phosphorylation state and MAPK activation remains unclear, although it has been demonstrated that the catalytic domain of Shp2 is required for sustained MAPK activation (Cross et al., 2002;
Hadari et al., 1998). In addition to the plasma membrane Ras activation pathway, Ras is activated on intracellular membranes of the Golgi apparatus and the endoplasmic reticulum upon RTK stimulation. Endomembrane Ras may be particularly important for sustained Ras activation. It has been described that endomembrane Ras activation requires the Src family kinases (SFKs) and recent studies have shown that Shp2 activates the SFKs.
In addition, Shp2 probably dephosphorylates the negative RTK regulator Sprouty (Spry) (Zhang et al., 2004).
There is a first hint that sustained activation of the MAPK by Shp2 results in mitogenic activation of the cell. Also the adapter Crk seems to be involved in mitogenic signalling (Hadari et al., 2001). And the protein kinases C λ and ζ have been shown to interact with FRS2 following FGF stimulation activating mitogenic signalling via the MAPK cascade (Lim et al., 1999).
Figure 1: Schematic representation of the mitogenic FGF signalling response.
3.4 The cell survival response
Recruitment and phosphorylation of the docking protein Gab1 to FRS2 associated Grb2 activates the cell survival pathway via the PI3 kinase. Gab1 binds constitutively to the SH3 domain of Grb2. Gab1 like FRS2 belongs to the family of scaffolding adapter proteins that are targeted to specific membrane lipids with their N-terminal pleckstrin homology domain (PH). It contains multiple
potential tyrosine, serine or threonine phosphorylation sites. The Drosophila homologue is Daughter of Sevenless (Dos) that was identified as potential substrate of Corkscrew (Csw), the Drosophila Shp2 ortholog. Soc1, the C.
elegans homologue was found in a screen for suppressors of hyperactive Egl- 15, the FGF receptor ortholog.
Interestingly, mammalian Gabs and Dos contain two Shp2/Csw binding sites (Gu and Neel, 2003).
Figure 2: Schematic representation of the cell survival FGF signalling complex.
3.5 The migratory response
There is evidence that activation of Shp2 is required to elicit the migratory response (Hadari et al., 2001; Rosario and Bircmeier, 2003). When the Shp2 binding sites of FRS2 are mutated, Grb2 can partially restore the chemotactic response. Thus, there are some redundancies in the function of the molecules.
It was also shown that the adapter Shb forms additionally to Shp2 a complex with Src and the focal adhesion kinase (FAK), a serine-threonine kinase that interacts with integrins. Integrins play a role in the regulation of focal adhesion assembly and disassembly, a process important during cell movement. Upon activation and phosphorylation of FAK in response to integrin mediated cell adhesion or cellular stimulation by agonists, FAK associates with Src. Shp2 promotes Src activation indirectly. It was shown that Shp2 dephosphorylates PAG , a transmembrane glycoprotein, thereby inhibiting the recruitment of Csk to PAG. Csk is responsible for the C-terminal inhibitory phosphorylation of Src (Zhang et al., 2004).
Dependent on the activity of RhoA, Src transits from the perinuclear region to the cell periphery requiring actin stress fibres to incorporate into focal adhesions or into smaller focal complexes at lamellipodia and filopodia. There, Src phosphorylates a number of focal adhesion components including paxillin, tensin and p130 Crk associated substrate CAS and also the p190RhoGAP (Holmqvist et al., 2003; Timpson et al., 2001). The role of FAK and Src is thought to be in the turnover of focal adhesions (Webb et al., 2004).
It has also been shown that FAK overexpression stimulates cellular migration, whereas removal of FAK decreases migration, which is associated with an increase in the number of focal adhesions and stress fibres (Ilic et al., 1995).
Interestingly, cells lacking Shp2 show hyperphosphorylation of FAK, increased numbers of focal adhesions at the cell periphery and a decreased migratory capacity similar to FAK-/- cells (Larsen et al., 2003).
In cell culture experiments, it was also shown that the urokinase type plasminogen activator (uPA), a proteolytic enzyme is upregulated upon FGF stimulation. uPA is involved in the extracellular matrix breakdown. Experiments in L6 myoblasts proposed that the Y463 and Y730 of FGFR1 could be essential for FGF2-mediated uPA induction (Boilly et al., 2000).
Figure 3: Schematic representation of the migratory FGF signalling response.
4. Negative regulation of FGF signaling
4.1 The ubiquitin ligase Cbl
FGF induced ternary complex formation of FRS2α, Grb2 and the RING type E3 ubiquitin ligase Cbl results in ubiquitination and degradation of FRS2α and FGFR. In vivo, Grb2/SOS and Grb2/Cbl complexes compete for binding to tyrosine-phosphorylated FRS2α.
In Cbl-deficient fibroblasts, the FGFR receptors are internalised in a normal manner, indicating that other mechanisms exist for downregulation of FGFR.
(Wong et al., 2002).
4.2 MAP kinase-mediated negative feedback mechanism
MAPK dependent phosphorylation of FRS2α on threonine residues results in reduced tyrosine phosphorylation of FRS2α upon FGF signalling and subsequent reduced recruitment of the Grb2/SOS complex and the phosphatase Shp2. As consequence the MAPK response and other downstream signalling cascades attenuate. Experiments using the Boyden chamber assay have shown that the mobility is enhanced of FRS2α-/- mouse embryonic fibroblasts (MEF) expressing a mutant form of FRS2α in which the threonine sites are mutated to valine. The mutant MEF cells expressing the threonine mutated FRS2α even grew faster than MEFs expressing the wild type FRS2α resulting in a transformed phenotype (Lax et al., 2002).
4.3 PLCγ as negative regulator of FGFR
Mice homozygous mutant for FGFR1 Y766F are viable but show a gain of function phenotype. Phosphorylation at Y766 (the binding site for PLCγ) has been implicated in the internalisation and degradation of the receptor. PKC a target of PLCγ has also been suggested to be involved in phosphorylation and downregulation of Src-family kinases, which are downstream targets of FGFR1 (Partanen et al., 1998).
4.4 Sprouty (Spry)
Another conserved inhibitor of RTK signalling is Sprouty (Spry). Spry was first identified in Drosophila as an inhibitor of FGF signalling during tracheal development (Hacohen et al., 1998) and subsequent studies have shown that it also inhibits signalling mediated by the EGFR during eye development and oogenesis in Drosophila (Casci et al., 1999; Kramer et al., 1999). Bnl signalling in the tip cells of the outgrowing tracheal branches regulates spry expression. In spry mutants, Aop/Yan, an ETS domain containing repressor of Bnl induced transcription, is degraded in an expanded domain. Subsequently, expanded domains of the transcription factor Pointed and the Drosophila Serum response factor homologue Blistered/DSRF can be observed, two transcriptional targets
that get expressed upon Bnl signalling. Genetic mosaic analysis has shown that Spry function is required in the tip cell and acts non autonomously to inhibit secondary and terminal branching by nearby stalk cells (for more details, see chapter about the development of the tracheal system in Drosophila) (Hacohen et al., 1998)
In contrast, during eye development, Spry acts cell autonomously as negative regulator of EGF signalling in R7 cells. Thus it is possible that Spry influences neighbouring cells in the tracheal system maybe indirectly (Casci et al., 1999;
Kramer et al., 1999).
In mammals, four Sproutys (Spry1-4) and three Sprouty-related EVH1 domain proteins (Spred1-3) have been identified. In both Drosophila and mammals, these proteins all consist of conserved carboxy-terminal cysteine rich domains and highly divergent amino-terminal domains. The four mammalian Sprys are known to be small phosphoproteins that form oligomers through their carboxy- terminal domains. The carboxy-terminal domains are also required for the translocation and for anchoring Spry to the plasma membrane through palmitoylation.
Sprys have been shown to act as inhibitors of the ERK signalling cascade but can also function to positively regulate this pathway. The slightly different activities are probably due to their different interaction partners, including Cbl, Grb2, Raf1, FRS2, caveolin-1, and the Drosophila RasGAP, Gap1 and Drk, the Grb2 homologue.
Spry1 and 2 get phosphorylated on Tyr 55 on the N-terminus, which serves as docking site for Cbl. The competitive binding of Cbl prevents the RTK from ubiquination, internalisation and the following degradation. Thus Spry acts as positive regulator resulting in prolonged and sustained signalling activity.
Finally, Cbl ubiquinates Spry and Spry gets degraded by the proteasome.
Upon FGF induced tyrosine phosphorylation, Spry translocates to the ruffling membrane region and binds to Grb2 at the same Tyr 55 as Cbl, thereby preventing recruitment of the Grb2/SOS complex to FRS2 or Shp2 and
inhibiting downstream events in the Ras/MAPK cascade (Hanafusa et al., 2002).
Another mechanism how Spry interferes with ERK activation is at the level of Raf. Spry2 and 4 physically associate with Raf1 through the highly conserved Raf binding motif (RBM) in the C-terminus thereby blocking phosphorylation of Ser338 on Raf1 that is essential for Raf activity.
In summary, Sprys have a positive and negative function residing in the N- terminal domain and a negative function in the C-terminal domain (Christofori, 2003).
4.5 Sef (similar expression of fgf genes)
The Sef protein is conserved across zebrafish, mouse and humans but not invertebrates. Sef encodes a putative transmembrane protein, with a signal and transmembrane domain. There is similarity to the intracellular region of mouse and human interleukin 17 (IL17) receptor. A putative tyrosine phosphorylation site juxtaposed to the transmembrane domain was shown to be important for FGFR1 and FGFR2 interaction. For the inhibitory effect of Sef, both the extracellular and the intracellular domain have to be functional. The intracellular domain of Sef interacts with the FGFR1 independent of FGFR1 tyrosine kinase activity.
In contrast to Sprouty, Sef seems to specifically repress FGF signalling (Furthauer et al., 2002; Kovalenko et al., 2003; Tsang et al., 2002).
Overexpression of mouse Sef in NIH3T3 cells does not inhibit ERK1/2 activation by PDGF, EGF for example.
Coimmunoprecipitation experiments have shown that Sef binds to the FGFR1 and mediates a reduction in tyrosine phosphorylation of FGFR1 and its immediate downstream target FRS2, thereby modulating multiple FGF- mediated signalling pathways as for example regulating the mitogenic response (Kovalenko et al., 2003).
II. Chemotactic cell movement
Cells move by locally extending filopodia (long extensions containing parallel actin bundles) and lamellipodia (dendritic actin network) being protrusions of the leading edge. Lamellipodia formation is driven by localised actin polymerisation under the control of Rho family of small GTPases in conjunction with loosening of the myosin thick filament network in the cortex to allow expansion to occur (Dormann and Weijer, 2003).
To gain traction, the cell has to attach to the substrate and make new contacts at the leading edge. The attachment sites mature to become focal adhesions that allow the cell to exert force upon its surroundings by actomyosin-dependent contraction to pull the cell body forward. Transmembrane adhesion receptors of the integrin family provide a link between the actin cytoskeleton and extracellular matrix (ECM) components and at focal adhesions, stress fibres, which are long bundles of filamentous F-actin, are linked to the integrins.
Sliding of myosin along actin filaments to mediate contraction requires specific phosphorylation of the myosin light-chain (MLC), which is carried out by the myosin light chain kinase (MLCK). The myosin light chain phosphatase (MLCP) opposes the action of MLCK and dephosphorylates MLC. MLCP is phosphorylated and inactivated by Rho kinase (Dossenbach et al.) downstream of Rho signalling to promote membrane ruffling and cell migration.
Finally, the cell contacts at the rear of the cell are released from the extracellular matrix and the membrane receptors are recycled from the rear to the front. Recent evidence indicates that Rho is required for tail retraction. Rho regulates also the phosphorylation of moesin, a plasma-membrane-actin- filament crosslinker (Larsen et al., 2003; Rogers et al., 2003).
Maximal cell migration occurs when cytoskeletal forces are in balance with adhesion. Under conditions in which the adhesive strength is too low, cells are unable to generate enough traction to move, whereas under conditions of high adhesiveness, cells are unable to break cell-substratum attachments.
Chemotactic cell movement involves the detection of gradients of signalling molecules. Interpretation of the signal results in polarisation of the actin and myosin cytoskeleton of the cell along the gradient resulting in directed cell movement (Dormann and Weijer, 2003).
It was also shown that an intact microtubule cytoskeleton is required to maintain the polarised distribution of actin-dependent distributions at the leading edge of migrating fibroblasts. Microtubules are arranged with their minus ends near the cell centre or anchored at the centrosome. The plus ends radiate towards the leading edge, where they display dynamic instability. Microtubules can grow along actin bundles. Behind the lamellum, microtubule breakage and depolymerization occurs as a result of the compressive forces of the converging actin to which they are bound.
One hypothesis is, that microtubule growth could promote local activity of Rac in the cell front to drive lamellipodia protrusion, focal complex formation and further microtubule growth. Microtubule shortening could activate RhoA behind the lamellum to drive actomyosin contraction and promote the stabilisation of the microtubules, to protect them from breakage.
A second hypothesis is, that microtubule-actin interactions orientate towards the leading edge, which could direct the delivery of signalling molecules or membrane components required for lamellipodia protrusion.
It has also been shown that during dynamic instability, microtubules specifically target focal contacts, and that the targeting frequency is inversely proportional to focal contact lifetime. Further evidence indicates that a kinesin microtubule motor may deliver a regulatory factor that promotes focal adhesion disassembly.
The Rho family of small GTPases also regulates the microtubules, for example RhoA mediates microtubules stabilisation. PAK kinases downstream of Rac1 also promote microtubule growth, probably by regulating the microtubule destabilising protein Op18/stathmin (Rodriguez et al., 2003).
Figure 4: Influence of phosphatases on cell migration. (A) Migration is initiated by protrusion of the leading edge and formation of new actin filaments. Rac induces lamellipodia and Cdc42 stimulates filopodia. Their actions are opposed by phosphatase and tensin homologue (PTEN). Rac activation is also opposed indirectly by protein tyrosine phosphatase (PTP)-PEST (for proline, glutamate, serine and threonine-rich domain). The actin-severing protein, actin depolymerizing factor (ADF)/cofilin, can be dephosphorylated by protein phosphatases (PP)1A and PP2A to stimulate migration. LIM (for LIN11, ISL1 and MEC3) kinase (LIMK), which is activated by Rho, phosphorylates ADF/cofilin. (B) Attachment at the leading edge occurs first with focal complexes, which develop into focal adhesions. Formation/turnover of focal adhesions is regulated by many phosphatases, some of which include: PP2A, SAP1 (for stomach-cancer-associated PTP) and PTP-PEST, all of which are generally inhibitory to migration; and PTP1B, SHP2 (Src-homology-2 domain-containing protein tyrosine phosphatase and PTP , all of which are generally stimulatory to migration. (C) Cell-body contraction results from forces generated through actomyosin interactions. Myosin light-chain (MLC) phosphatase (MLCP) dephosphorylates MLC to inhibit migration and oppose the action of MLC kinase (MLCK). Rho kinase (Dossenbach et al.) phosphorylates MLCK and inhibits its activity. (D) Rho stimulates tail retraction. Rho also phosphorylates the actin-binding-protein moesin, which is dephosphorylated at the rear of the cell, before rear release. Moesin is dephosphorylated by a
1. Molecular requirements for actin-based lamellipodia and filopodia formation
New actin filaments are nucleated by the Arp2/3 complex and grow in a polarised fashion with the fast growing barbed ends oriented toward the leading edge. Arp2/3 is activated by the WASP (Wiskott-Aldrich Syndrom protein) and SCAR family of proteins, which are in turn activated through small G proteins.
Cortactin also binds and activates Arp2/3 complex and at the same time stabilises branches.
Fed by monomeric actin-profilin from the subunit pool, new branches grow rapidly and push the membrane forward (for the exact process see Fig.1).
Profilin allows elongation of barbed ends of filaments, blocks binding to the pointed end and inhibits spontaneous nucleation of actin filaments. Capping by the Capping protein or Gelsolin, another capping protein, limits the length of the growing branches, since short filaments are stiffer and more effective at pushing on the membrane.
The Ena/VASP proteins have been shown to interact with the barbed ends of actin filaments, shielding them from the activity of capping protein while supporting filament elongation which is important for the cell to create more filopodia like structures to sense guidance cues in the environment (Bear et al., 2002). Toward the rear of the lamellipodia actin filaments become debranched, severed and depolymerised by Cofilin-like proteins (ADF/Cofilin) and the released monomeric actin or globular G-actin is recycled into polymer at the leading edge (Petit et al., 2002; Rogers et al., 2003). Furthermore, Cofilin is required at the initiation of protrusion for barbed end mediated actin assembly at the leading edge and thereby increases levels of Arp2/3-mediated polymerisation (Larsen et al., 2003; Pollard and Borisy, 2003).
Arp2/3 has not been detected in filopodia in which actin is arranged in bundles rather than in a branched array as in lamellipodia.
The question remains how external stimuli are converted into signals that regulate nucleation-promoting factors. The strong autoinhibition of WASP is overcome by signalling molecules including Rho family GTPases (see next chapter), PIP2, Profilin, Grb2 and Nck. Requiring several signalling inputs of
different natures is attractive because it makes the mechanism of actin polymerisation sensitive to coincident signals (Carlier et al., 2003).
Figure 5: Schematic diagram representing the reactions involved in building a branched filament array at the leading edge. (A) Activation and targeting of N-WASP at the plasma membrane by signalling molecules. (B) Formation of the ternary branching complex G-actin–N- WASP–Arp2/3. (C,D) Association of the branching complex with a nucleus or filament barbed end. (E) Growth of the branched filament from G-actin and profilin-actin. (F,G,H) Regulation of actin dynamics in lamellipodium by Actin Depolymerising Factor and capping proteins (represented by gelsolin, which makes a 1:2 complex with G-actin). (I) Sequestration of actin by thymosinb4. (J) Recycling of G-actin by nucleotide exchange (Carlier et al., 2003).
2. Role of the small Rho GTPases in cytoskeletal reorganization
Members of the Rho (Ras homologous) family of GTPases are major components regulating changes in cell morphology and reorganization of the cytoskeleton in response to external stimuli (see also chapter II). Each Rho- family GTPase contains a membrane-attachment domain, a GTP/GDP binding domain and an effector loop. These molecules cycle between inactive GDP- bound and active GTP-bound forms. The cycling is regulated by guanine nucleotide exchange factors (GEFs), which exchange GDP for GTP, and GTPase-activating proteins (GAPs), which induce the hydrolysis of, bound GTP to GDP. Guanine nucleotide dissociation inhibitors (GDIs) prevent dissociation of GDP and the hydrolysis of bound GTP thereby preventing activation.
In fibroblasts, RhoA is implicated in the formation of actin stress fibres and focal adhesions, Rac1 stimulates membrane ruffling, lamellipodia, focal complex formation and is present at cell junctions, and Cdc42 participate in the formation of filopodia, cell rounding and the loss of actin stress fibres (Van Aelst and D'Souza-Schorey, 1997).
Figure 6: GTPase cascades involved in cytoskeleton organization in fibroblasts.
Various extracellular stimuli trigger the activationof Cdc42, Rac, and Rho GTPases and elicit specific short-term responses such as the formation of filopodia, lamellipodia, and stress fibres, respectively. Moreover, Cdc42 appears to activateRac, which in turn activates Rho. The direct links between theseGTPases remain to be clarified. Different agonists can activate the Rho GTPases independently. The mechanism by which theseagonists activate Rho GTPases may involve GEFs, GAPs, or GDIs (Van Aelst and D'Souza-Schorey, 1997).
The role of Rho family GTPases is not fully understood, numerous studies using mutants as well as dominant negative and constitutive activated forms of these GTPases support their importance in cell migration in vitro and in vivo (Petit et al., 2002).
One of the targets of Rac1 is the phosphatidylinositol-4-phosphate-5-kinase (PIP 5-kinase). Through the increase of PIP2 levels, PIP5-kinase regulates the function of many actin-associated proteins.
Rac1 mediates initiation of Cadherin-mediated cell-cell adhesion in a PI3- kinase-dependent manner. Rac1 is specifically expressed in initiating areas of contact where lamellipodia are formed, whereas E-cadherin gradually accumulates along the entire contact length. The regulatory subunit of PI3K associates with the E-cadherin/catenin complex and the Rac1 GEF Tiam1, which promotes E-cadherin based cell-cell adhesion, is regulated by PI3K activity (Ehrlich et al., 2002).
Another target of Rac1 is POR1 that specifically interacts with Rac1 in GTP- dependent manner. Deletion mutants of POR1 inhibit the induction of membrane ruffles by RacV12, a constitutive active form of Rac. POR1 interacts with the GTPase Arf6 that has been shown to elicit cytoskeletal rearrangements at the cell surface. But experiments have shown that Arf6 and Rac function on distinct signalling pathways to mediate cytoskeletal reorganisation (Van Aelst and D'Souza-Schorey, 1997).
WASP is regulated by Cdc42. WASP contains multiple domains that interact with different signalling molecules, phosphoinositides and components of the machinery required for actin polymerisation, such as actin monomers and the Arp2/3 complex. Upon activation by GTP-Cdc42, the intramolecular inhibition of
WASP is released and permits the binding of WASP to the Arp2/3 complex.
This complex formation activates the actin-nucleation thereby locally increasing actin polymerisation.
Contraction and stabilisation of stress fibres is obtained when phosphorylated myosin light chain (MLC) is associated with actin. Phosphorylation of MLC induces a conformational change in myosin, thereby increasing its binding to actin filaments and subsequently the formation of stress fibres. MLC phosphorylation is regulated by the opposing effect of MLC-kinase (MLCK) and MLC-phosphatase. Some targets of Rho proteins, for example PAK, a serine/threonine kinase, exert their effects by regulating the phosphorylation state of MLC. PAK is required for the disassembly of focal adhesions and promotes lamellipodia formation and membrane ruffling. Moreover PAK activation leads to a decrease in MLC phosphorylation through phosphorylation of MLCK, thus destabilising actin stress fibres.
In contrast, the Rho associated serine/threonine kinases ROKα/β that are targets of the Rho GTPase, phosphorylate MLC and inactivate the MLC- phosphatase thus increasing actomyosin assembly.
III. Developmental role of FGF signalling in Vertebrates
1. Tumorogenesis
Several FGFs were discovered because of their oncogenic potential. For example in human pancreatic cancer and in breast cancer, in lung carcinomas or glioma, FGFs are over expressed.
Some tumours are mainly spread through the lymph, others mainly through the blood stream. A positive correlation between the vascular density of the primary tumour and the number of metastases formed has been reported for several cancers (Klingenberg Dissertation).
Several antagonists of FGF signalling that are of pharmaceutical interest were tested for example antisense molecules, neutralising antibodies, dominant- negative FGFRs or soluble FGF receptors.
Blockade of the FGF pathway in mice by FGF2 or FGFR1 antisense molecules or by FGF2 neutralising antibody inhibits tumour angiogenesis and growth.
In mice, tumour growth can also be inhibited by the adenoviral expression of soluble FGFR1 and VEGFR1. FGFR1 decreases the number of capillaries in tumours suggesting that growth is inhibited through an angiogenesis dependent mechanism, and the coexpression of soluble VEGFR1 has an additive effect in the inhibition of tumour growth (Auguste et al., 2003).
2. FGF-mediated movement of mesoderm cells during gastrulation
Gastrulation is the process that establishes the three-layered body plan, the ectoderm, mesoderm and endoderm of the embryo and is one of the main morphogenetic events that shape the early embryo.
In mammalian and avian embryos first the embryo consists of two layers, the epiblast and the hypoblast. The epiblast gives rise to embryonic and
extraembryonic structures whereas the hypoblast gives rise to extraembryonic structures only. Gastrulation starts with the formation of the primitive streak. The anterior part of the streak is known as Hensen’s node. During gastrulation, cells move towards the primitive streak where they undergo an FGF-mediated
“epithelial to mesenchymal transition” (EMT) that requires downregulation of E- cadherin, a process probably regulated by FGFR1. Chimeric analysis of FGFR1-/- mice has shown that FGFR1 signalling is required for EMT to occur in mouse embryos and that FGFR1-/- deficient cells have strong migration defects (Ciruna and Rossant, 2001). FGFR1 signalling is required for the expression of mSnail that downregulates E-cadherin. Mouse embryos homozygous mutant for mSnail die late during gastrulation with mesodermal cells retaining epithelial characteristics including the expression of E-cadherin. Downregulation of E- cadherin at the primitive streak not only regulates the EMT and migration of mesoderm progenitor cells, but also permits the rapid accumulation of cytosolic β-catenin levels in response to localised Wnt signalling and Wnt target genes are expressed, as for example Brachyury and T box genes that specify mesodermal cell fates (Ciruna and Rossant, 2001).
The protein tyrosine phosphatase Shp2 seems to be required to positively transmit the FGFR1 signal. Shp2 mutant cells in chimeric mice accumulate within the primitive streak because of the incapability of the Shp2-deficient cells to undergo EMT, which involves changes in cell shape, adhesion or migration.
These processes are essential for the cells to exit the streak and to contribute to the mesoderm. The behaviour of Shp2 mutant cells looks very similar to the one of FGFR1-/- cells in chimeras suggesting a specific role of Shp2 downstream of FGFR1 to induce chemotaxis (Saxton and Pawson, 1999).
After passing the streak, the cells move out as individual cells into the space between the epiblast and the hypoblast to form the axial and lateral mesodermal, the definitive endoderm and extra-embryonic structures.
The phenotypes of knockout mice already suggested that FGFs are involved in cell movement during gastrulation, since FGF8 and FGFR1 mutants showed severe defects in cell migration at the primitive streak stages (Yang et al., 2002).
On the basis of the expression pattern of FGF8 and FGF4 (both are expressed in the streak and partially overlapping) it has been shown that mesodermal cells are attracted by sources of FGF4 and repelled by FGF8. It was proposed that cells move away for the primitive streak as a result of repulsion by FGF8 and that cells emerging from the anterior streak are attracted back in towards the midline to form somites and lateral plate mesoderm after regression of the node, by a FGF4 signal from the forming notochord. Cells leaving the posterior streak are attracted by a source of unknown chemoattractant originating from the boundary region between the area opaca and area pellucida to form extraembryonic structures. These cells never sense the attractive influence of the forming notochord and somites (Dormann and Weijer, 2003).
Figure 7: FGF expression directs cell movement in chick embryos. Movement of cells emerging from the primitive streak repelled by FGF8 (orange) produced in the streak.
Mesoderm cells are attracted back in towards the midline to form somites and lateral plate mesoderm (blue arrows) by FGF4 (green) produced by the forming head-process. Cells emerging from the posterior streak (black arrows) are attracted by a signal emitted from the boundary of the embryo (Dormann and Weijer, 2003).
3. Angiogenesis
Angiogenesis is the formation of new blood vessels from pre-existing vessels.
Angiogenic factors, as the FGFs and the Vascular Endothelial Growth Factors (VEGFs), stimulate endothelial cells to secrete several proteases and
plasminogen activators resulting in the degradation of the vessel basement membrane, which allows the cells to invade the surrounding matrix. The cells migrate, proliferate and differentiate to form a new lumen-containing vessel.
Finally, the endothelial cells deposit a new basement membrane and secrete growth factors such as the platelet-derived growth factor (PDGF), which attract supporting cells, the pericytes, ensuring the stability of the new vessels. This complex process involves other factors, as the angiopoietins and ephrins that act on specific receptors to regulate vessel stability (Cross and Claesson- Welsh, 2001).
FGF2 (also called basic FGF) was the first angiogenic factor identified and the FGFR1 is the main FGFR expressed in endothelial cells in vitro and has also been detected in activated endothelial cells in vivo. In capillary endothelial cell lines, stimulation of FGFR1 induces proliferation, migration, protease production and tubular morphogenesis, whereas activation of FGFR2 increased motility only.
Studies in FGF2-/- mice have shown, that endothelial cells from these mice have a defect in cell migration that can be compensated by the addition of exogenous FGF2 (Auguste et al., 2003).
FGFs probably regulate vascular morphogenesis indirectly by inducing secondary angiogenesis regulators such as VEGF.
Disruption of FGF or FGFR genes was not very informative so far with regard to vessel formation in vivo. FGF1-/- or FGF2-/- mice (single or double knockouts) do not present a defective vascular phenotype during development. This suggests a functional redundancy or a non-essential role of FGFs in vasculogenesis or angiogenesis. Mice with gene knockouts of FGFR1 or FGFR2 yield embryos arrested in their development before the onset of vascularization, because of the lack of mesoderm inducing signals.
But overexpression of FGF1 or FGF2 in mice hearts lead to an increase of vessel density. And transgenic mice that overexpress specifically a dominant- negative FGFR1, truncated at the C-terminal kinase domain, in the retinal pigmented epithelium in the developing eye, show a poorly branched vascular bed in the choroid and an avascular neonatal retina (Auguste et al., 2003).
4. Development of the mammalian lung
4.1 The morphology of the lung
Figure 8: Human lung. (B) In white, preparation of the lung of an adult human using acryl polyester to fill in the airways. View from behind. The left lung has been filled less than the right half. Courtesy of H. Kurz, Anatomical Museum, University of Basel, Switzerland. In red, the descending aorta is visible (Affolter and Shilo, 2000).
The mammalian lung evolved as a system of branched conduits for air and blood coupled to a vast network of alveolar structures designed for gas exchange. In the developing respiratory system, airway branching is a prenatal event and formation of the alveoli spans pre- and postnatal life.
Pulmonary branching is reproducible in its spatial pattern. This suggests that the pattern of respiratory morphogenesis in the first 16 airway generations is genetically predetermined. Between the level of 16 and 23 generations, in which the alveoli are formed, branching appears more randomly.
Primordial lung buds originate as outpouchings of the primitive foregut endoderm and reiterated budding and branching of these tubules generate the airway tree. As the lung develops, vascular and airway components intermingle at the distal end of this tree to form the future alveolar-capillary barrier.
Endodermal cells of the ventral foregut form the epithelial lining of the tube.
Lateral plate mesodermal cells migrate and condense around the endoderm to form the mesenchyme.
Left-right differences in pattern are first seen when secondary buds form (see Fig.1).
The genes lefty-1, lefty-2 and nodal are expressed on the left side of mouse embryos and are implicated in the determination of left-right laterality (Warburton et al., 2000). These differences continue to develop as airways
undergo branching and are assembled into lobes, which result in more lobes in the right lung than in the left (mouse and human).
4.2 Cell fate determination
The homeodomain protein Nkx2.1 is essential for the complete induction of the lung branching morphogenesis. In the absence of Nkx2.1, dorso-ventral separation of the trachea from the esophagus, as well as lung branching morphogenensis and epithelial cell lineage determination at an early stage, prior to the specification to peripherial lung cell phenotypes is arrested. The hepatocyte nuclear family (HNF) of transcription factors cooperate with Nkx2.1 to determine pulmonary epithelial cell fate. HNF-3 activates transcription of Nkx2.1 in respiratory epithelial cells. Another important factor is GATA-6, a member of the GATA family of zinc finger transcription factors, induces differentiation of primitive foregut endoderm into respiratory epithelial cell lineages by interacting with HNF3β, Nkx2.1 and GATA family members (Warburton et al., 2000).
4.3 The branching process
A large amount of factors are involved in the branching process. The most relevant will be discussed. In the lung branching process, localised dynamic FGF10 expression in the distal mesenchyme induces chemoattraction and epithelial cell proliferation. Little is known, how specific clusters of mesenchymal cells are determined in a stereotyped fashion to secrete FGF10 and stimulate bud outgrowth.
Experiments have demonstrated directional chemotactic bud outgrowth in mesenchymal free epithelial cell cultures or whole lung cultures toward an implanted FGF10 soaked heparin bead. In FGF10 knockout mice, induction of primary bud outgrowth is disrupted and the mice have a blunt ended tracheal tube (Warburton et al., 2000). Other experiments have shown that in mesenchyme free lung epithelial cultures differential cell proliferation does not seem to be the initial event that triggers lung bud induction (Cardoso, 2001).
